Full-stack replacement for a legacy .NET trading platform. Built from scratch with 157 React components, 127 REST endpoints, real-time order book with matching engine, and 119.7 GB data migration from SQL Server to PostgreSQL.

ReactExpressTypeScriptPostgreSQLRedis

View Project

email-triage

[09:15:22] Polling inbox... 5 unread

Analyzing: "Re: Project deliverables"

→ Client Request (0.94) | Priority: High

→ Skill match: Project Management

✓ Task created → Assigned to @pm-lead

Analyzing: "Weekly Newsletter"

→ Marketing (0.99)

→ Rule: skip-newsletters

⊘ Skipped (auto-archive)

Analyzing: "Invoice #4521 attached"

→ Finance Request (0.91)

→ Skill match: Accounting

✓ Task created → Assigned to @ops-team

Analyzing: "Quick question about API"

→ Technical Support (0.88)

✓ Task created → Assigned to @pm-lead

[09:15:31] 3 tasks created | 1 skipped

Live

Claude AI

Email Triage Assistant

AI-powered email classification system using Claude that monitors Office 365, applies intelligent routing rules, and automatically creates Asana tasks. Running 24/7 in production on personal infrastructure.

Claude AINode.jsMicrosoft GraphAsana APISQLite

View Project

memory-mcp

[09:12:01] memory-mcp v2.0 online

[09:12:01] ✓ PostgreSQL + pgvector (HNSW)

[09:12:02] ✓ Neo4j graph (891 entities)

[09:12:02] ✓ Redis cache connected

[09:12:03] ✓ Embedding model loaded (384-dim)

[09:12:15] STORE scope:project:vaultkeeper

→ Embedded (12ms) │ 3 entities extracted

→ Stored #247 │ graph updated

[09:12:22] SEARCH "auth token rotation"

→ hybrid + rerank │ 23 candidates

→ #1 (0.94) "JWT refresh token strategy"

→ #2 (0.87) "Redis session management"

[09:12:30] TRAVERSE entity:"vaultkeeper"

→ 12 connections │ 4 projects │ 3 decisions

[09:13:00] PROMOTE dry_run=false

→ 247 scored │ 14 pinned │ 3 archived

── 247 memories │ 891 entities │ 12 tools ──

Live

AI Memory System

A 5-layer distributed memory architecture that gives AI coding assistants persistent recall, relationship reasoning, and automated lifecycle management.

Node.jsTypeScriptPostgreSQL+3

What industry are you focused on? I'll find the communities where your audience is talking.

Commercial real estate — looking for tenant pain points and emerging market trends.

Found 8 relevant communities. Here are the top matches:

r/CommercialREr/realestateinvestingr/CREr/PropertyManagementr/RealEstate

Top 3 look good. What analysis can you run?

I can generate 5 report types — pain points, opportunities, trends, comprehensive, or dual-perspective. Which profile fits you?

Content CreatorConsultant

Discovering subreddits...

Describe your business...

↑

Live

Reddit Intelligence Platform

AI-powered market research tool that turns Reddit into structured intelligence. A conversational onboarding experience builds a user profile and maps relevant communities, then generates targeted research briefs across five analysis types.

Next.jsFastAPIClaude AI+2

product-sync

[05:00:01] sync-service started (v2.4)

[05:00:01] Connecting to WordPress API...

✓ Connected | WooCommerce v3

[05:00:03] Fetching product catalog...

✓ 384 products retrieved

[05:00:04] Checking description cache...

→ 341 cached (hash match)

→ 43 new/modified products

[05:00:05] Generating AI descriptions...

→ Model: claude-3-haiku

→ 43 descriptions queued

✓ 43/43 generated ($0.27)

[05:00:14] Updating Google Sheets...

→ 384 rows synced

→ 43 descriptions updated

✓ Spreadsheet updated

[05:00:16] Sending notification...

✓ Email sent to 3 stakeholders

── Weekly Sync Complete ──

Products: 384 │ New: 43 │ Cache: 89%

AI cost: $0.27 │ Duration: 22s

Live

Product Sync Pipeline

Automated weekly sync of 100+ products from WordPress to Google Sheets with AI-generated marketing descriptions and branded stakeholder notifications.

PythonClaude AIGoogle Sheets API+2

Performance by Territory

Top Locations

Alerts

Station #2104 — low rating

Station #1832 — no response

14 reviews pending reply

Live

Sunoco GMB Analytics Dashboard

Real-time analytics platform for 2000+ fuel station locations with role-based access control, territory management, and performance leaderboards built on React and Supabase.

ReactSupabaseTypeScript+2

brand-engine

[08:30:01] brand-engine started (v2.1)

[08:30:01] Loading master.pptx...

✓ 292 slides parsed | 1,847 shapes indexed

[08:30:03] Processing 7 retail partners...

▸ Partner 1/7

→ Logo swap: 34 placements

→ Color remap: LAB ΔE < 2.0 (142 fills)

→ Text tokens: 68 replacements

✓ 51 modules generated

▸ Partner 2/7

→ Logo swap: 34 placements

→ Color remap: LAB ΔE < 2.0 (142 fills)

→ Text tokens: 71 replacements

✓ 51 modules generated

→ ...5 more partners

── Batch Complete ──

Partners: 7 │ Files: 357 │ Slides: 2,044

Duration: 8m 12s │ Errors: 0

Live

Enterprise Branding Automation

Automated PowerPoint rebranding system that transforms 292-slide master presentations into 51 brand-specific training modules in 5-10 minutes - a 97% reduction from manual processing.

Pythonpython-pptxPillow+2